Utility of the diffusion-weighted imaging for activity evaluation in Crohn's disease patients underwent magnetic resonance enterography

Abstract

Background: Cross-sectional imaging techniques as magnetic resonance enterography (MRE) may offer additional information on transmural inflammation, stricturing and fistulising complications in Crohn's disease (CD). The purpose of our study was to evaluate the diagnostic accuracy of Magnetic Resonance Imaging (MRI) combined with Diffusion-weighted Imaging (DWI) and MRE for determination of inflammation in small bowel CD.

Methods: MR imaging examination was performed with a GE Signa EXCITE 3.0 T MRI scanner. The optimal b value in DWI with a learning cohort of patients was determined. The diagnostic accuracy for active lesions and disease activity were accessed by MRE combined with DWI.

Results: The b value 800 s/mm(2) group showed the highest diagnostic sensitivity (74.19%) for diagnostic assessment of active Crohn's lesions on DWI. MRE combined with DWI showed the highest sensitivity (93.55%), specificity (89.47%) and diagnostic accuracy (92%) compared with MRE or DWI alone. The segmental MR score (MR-score-S) showed a significantly positive correlation with the Capsule Endoscopy Crohn's Disease Activity Index Score (CECDAI-S) (r = 0.717, p < 0.01). The total MR score (MR-score-T) showed significant association with C-reactive protein (CRP) (r = 0.445, p = 0.019) and erythrocyte sedimentation rate (ESR) (r = 0.688, p < 0.01).

Conclusions: MRE combined with DWI improves the diagnostic accuracy for active lesions and correlates the endoscopic disease activity. MRE with DWI could represent a non-invasive tool in assessing active inflammation in CD.

Figure 1

Determine the optimal b value in DWI with a learning cohort of patients. The mean CNR and SNR of four b value sequences. The unit of b value is s/mm²(A). The ADC values of four b value sequences. The unit of ADC is 10⁻³ mm²/s (B).

Figure 2

Qualitative image quality comparison among four b values. DWI images were obtained from a 30-year-old man. The Inflammatory lesion of approximal ileum was detected by endoscopy. When b = 800 s/mm² was adopted, the lesion was demonstrated clearly as well as normal small intestinal wall, enlarged lymph nodes and abdominal wall structure (A). When b = 1500 s/mm² was adopted, normal small intestinal wall could barely be identified (B) and when b = 2000 s/mm² or b = 2500 s/mm² was adopted (C and D), even abdominal wall structure became vague. The enlarged lymph nodes could not be identified when b = 2500 s/mm² was adopted (D).

Figure 3

Diagnostic effect of different b value groups. In a 28-year-old man, high signal intensity in approximal ileum could be detected when b value was 800 s/mm²(A) or 1500 s/mm²(B) was chosen. However, when b value was 2000 s/mm² or 2500 s/mm²(C and D) was selected, the high signal was apparently depressed and the lesion was missed.

Figure 4

The combination of MRE and DWI could improve diagnostic sensitivity and specificity. A 34-year-old female patient suffered from active CD with multi-segmental lesions in ileum (A, B and E). On LAVA image (A) abnormal enhancement could be detected in pelvis (white arrow), which was confirmed by enteroscopy (E), While on DWI (B), no abnormal high signal intensity lesion was found in the same cross-sectional slice. A 42-year-old male patient suffered from active CD (C, D and F). On LAVA image, suspicious local bowel wall thickening (curved arrow-head) was detected in distal ileum (C), while on DWI (D), no high signal intensity was found in the same cross-sectional slice, which was confirmed by enteroscopy (F).

Figure 5

The combination of MRE and DWI could improve diagnostic sensitivity and specificity. A thirty-three male patient suffered from active CD. On LAVA image the jejunum (right upper abdomen) was not distended very well and it was difficult to decide whether it had inflammatory lesion because of high signal intensity both before (A) and after (B) contrast injection. While on DWI (C), abnormal high signal intensity was found in the same cross-sectional slice indicating the inflammatory lesion, which was confirmed by enteroscopy.

Figure 6

The correlation between MR score and CECDAI score. Both MR-score-S and MR-score-T showed a significantly positive correlation with CECDAI-S (r = 0.717, p < 0.01) and CECDAI-T (r = 0.535, p < 0.01).

Figure 7

The ROC curve of predicting CD activity. The ROC curve indicated that bowel wall enhancement was the optimal predictor of inflammation (AUROC = 0.87). Besides, DWI hyper-intensity (AUROC = 0.83) and bowel wall thickening (AUROC = 0.80) could also predict active inflammation pretty well. At the same time, the layer differentiation of DWI hyper-intensity and differentiation between the mucosaesubmucosa complex and the muscularis propria presented the perfect diagnostic specificity.